The proper measurement (integration) of the peak areas in analytical chromatograms often calls for the analytical chemist to make judgements regarding accurate and consistent integration. As with many processes calling for human judgement there is the potential for this judgement to be influenced by non-objective criteria.

This article discusses the application of objective and scientifically sound processes to yield chromatographic integration that can withstand scientific and regulatory scrutiny.

1. Introduction

High Performance Liquid Chromatography (HPLC) and Gas Chromatography (GC) are two of the most routinely used techniques in the modern analytical laboratory. The specificity, sensitivity and flexibility of the two techniques make them readily applicable to the determination of an enormous range of analytes in a wide range of sample matrices. The raw output from both techniques is a graph of signal strength verses time, usually called a chromatogram. This consists of a series of peaks which represent the component substances of a sample. The area of the peaks is usually proportional to the amount of the respective substance in the sample.

In order to quantitate the analytes in the sample it is necessary to determine the area of the peak; a process known as integration. Modern HPLC and GC instruments are invariably interfaced with computerised chromatographic data capture and processing systems which are capable of performing the integration process automatically. The role of the analytical chemist is to select appropriate values for the parameters (such as: slope sensitivity, noise threshold, peak width, area threshold, and bunching factor and skim ratio) which are used by the processing software to define the respective chromatographic peaks.

Related: Train Your Staff in Laboratory Data Integrity

2. The Regulatory Perspective

As assigning the most appropriate chromatographic integration parameters requires a certain amount of judgement on the part of the analytical chemist, this presents the following issues from the quality assurance perspective:

• Assuring that the integration parameters are used, and therefore the peak integration itself is performed in a consistent way, and
• Assuring that that the integration of peaks, (which when integrated and calibrated against a chromatogram of a standard solution of the respective analyte, may yield an out of specification result), is not manipulated in such a way as to give a result that conforms to specification. In other words, using the integration parameters to manipulate the peak area, in order to obtain a passing result from a peak that would otherwise produce a result that would fail to meet specifications. This is sometimes called integrating, or testing, in to compliance.

In the heavily regulated pharmaceutical industry, the proper integration of chromatographic peaks and the processes used to achieve accurate peak integration attracts significant scrutiny during regulatory audits, and the assessment process for marketing authorisations for pharmaceutical products. The expectation of the regulatory agencies is that, the assignment of integration parameters and the integration of chromatographic peaks will be controlled by a standard operating procedure (SOP) or similar document, and follow a scientifically sound process.

A review of warning letters issued by the US Food and Drug Administration (FDA) revels that

• “Changing the integration parameters until the integration looks good” is unacceptable1. Indeed multiple re-integrations of chromatograms without explanation or justification are considered indicative of attempts to manipulate the respective peak areas to obtain passing results.
• Once a method has been validated and is being used for the routine quality control release of pharmaceutical products, the same integration parameters should be used for each chromatographic run, and indeed the current expectation seems to be that chromatographic methods should be ‘secured’ to assure the same integration parameters are always being used2. This, however, raises the question if the chromatographic method cannot be changed, how should atypical situations (such as extra peaks) be handled? The author does not advocate ‘locking’ methods to prevent changing parameters, but instead recommends including suitable chromatographic integration parameters in the approved written method, and allowing the analytical chemist to make appropriate adjustments, should this be necessary in unusual circumstances. In these situations it may be necessary to initiate an investigation, in order to determine the root cause of the unusual event.

3. Establishing Scientifically Sound Processes for the Integration of Chromatographic Peaks

So how should a laboratory establish and control used to integrate chromatograms?

3.1 The Integration Process

In order to establish control of the integration process it is necessary to understand how chromatographic data processing programs perform this task. The precise processes used to detect and define a peak in a chromatogram will depend, to some extent, on the algorithms used by the processing software. The discussion here will attempt to provide a generic overview, however it is impossible to provide a comprehensive discussion which covers all chromatographic data processing systems.

Essentially chromatogram integration consists of the following component events:

1. Defining the Initial Baseline. This is typically done by establishing an initial baseline by taking the first data point as a tentative baseline point. The integrator then attempts to redefine this initial baseline from an average of the input signal. The time period over which this averaging takes place is determined by the Peak Width or Bunching Factor Functions. If a redefined baseline is not established, the first data point is taken as the initial baseline point.

2. Tracking the Baseline. As the integrator continues to sample the input data, at a rate determined by the Peak Width Function, it considers each data point as a potential baseline point. A baseline envelope or threshold is determined, either from the slope (determined from the first derivative) and curvature (determined from the Second Derivative) of the baseline, or from entered parameters such as Noise Threshold. This baseline envelope or threshold can be considered as a cone with its tip at the current data point (Figure 1 A). As new data points are received, the cone moves forward until a break-out occurs (Figure 1 B). To be accepted as a baseline point, a data point must comply with the following criteria:

• It must lie within the defined baseline envelope or threshold
• The curvature of the baseline must be below a threshold value determined by the Slope Sensitivity Function or
• The cumulative sum must not eventually exceed an entered Area Threshold value

The baseline is continuously reset at a rate determined by the Peak Width or Bunching Factor Functions, and is periodically reset to compensate for baseline drift until a peak up-slope is detected.

3. Identifying the Peak

A peak is potentially identified when the potential baseline points lie outside the baseline envelope or threshold, and the baseline curvature exceeds a set value determined by the Slope Sensitivity Function. If this continues an up-slope of the peak is recognised. The potential start of a peak is recognised when the following criteria are met:

• The slope and curvature exceed the limit established by the baseline envelope or threshold, (Figure 1 B) or
• The difference between successive data points exceeds the Noise Threshold, a potential peak is identified starting

Following the identification of a potential peak, the start of the peak is defined if the following criteria are met:

• The slope remains above the limit established by the baseline envelope or threshold, (Figure 1 B) or
• The cumulated area sum exceeds the entered Area Threshold value,
• The curvature becomes negative forming the front inflection point (Figure 1 B)

The apex of the peak is recognised when the

• Slope passes through zero and becomes negative (Figure 1 C),
• Curvature becomes positive, or
• The data point is lower that the preceding one and the cumulative sum of the subsequent data points exceeds 2/3 Area Threshold, (Figure 1 [(C0-C1) + (C0-C2) + (C0-C3)])

The potential end of the peak is recognised when the

• Slope and curvature are within the limit established by the baseline envelope, or
• The differences between two consecutive data points are less than 0.5 of the entered Noise Threshold value

The end of the peak is defined if

• Slope and curvature remain within the limit established by the baseline envelope (Figure 1 D)

3.2 Fundamental Requirements and Policies

The first step in establishing scientifically sound integration processes is to define what is considered to be a peak, and what is considered to constitute proper and accurate integration of chromatographic peaks. The principle criterion should be the whole peak and only the whole peak should be integrated.

This can be illustrated, in an SOP, using diagrams of acceptable and unacceptable peak integration. Such as shown in Figure 2, where peak A represents an acceptably integrated peak, whereas peak B represents unacceptable peak integration, as the entire peak is not being integrated; and peak C is also unacceptable, as additional area which is not part of the peak is being included in the integrated peak area.

If quantitation of sample amounts is desired, which is usually achieved by comparing the peak area from a sample chromatogram with the peak area from a chromatogram of a standard solution of know concentration, It is particularly important that the chromatograms are integrated in an identical manner.

It is especially important that the laboratory establishes a clear policy that manipulating the area of a peak in order to achieve some predetermined outcome, such as a numerical result meeting some specification, is unacceptable.

3.3 Controlling the Integration Process

The effect of changing the chromatographic integration parameters on the final results can be investigated during the accuracy and precision phases of the method validation effort. Using the raw data obtained during the recovery experiments, it is suggested the following integration parameters (depending on those used by your processing software):

• Slope sensitivity
• Peak width
• Bunching factor
• Area reject
• Noise threshold
• Area threshold

Should be changed in a systematic manner and the effect on the calculated recoveries should be documented and assessed, for example as shown in Figure 3. This shows the variation of percent recovery as a function of slope sensitivity and peak width for some simulated data. Using a common acceptance specification of 98.0 – 102.0 %, shown in green, is achieved with a peak width of between 0.3 and 0.5 minutes and a slope sensitivity factor of between 10 and 75.

An optimum peak width – slope sensitivity combination is 0.4 minutes and 45, respectively. Using data gathered from studies like this, during the validation phase of a method’s life cycle, it is possible to develop a scientific rational for the selection of the most appropriate method integration parameters.

3.4 Unresolved Peaks

For the most accurate integration of chromatographic peaks it is necessary that all of the peaks are fully separated. If some of the peaks that require accurate quantitative determination are not fully separated a significant attempt at achieving separation, by optimising the chromatography, is called for. Obtaining numerical data from unresolved peaks should only be acceptable, if after a substantial effort, peak separation is unsuccessful.

If quantitative data must be obtained from unseparated peaks, the laboratory needs to establish clear policies as to how such peaks should be integrated. This should include policies on when it is acceptable to use different functions for integrating unresolved peaks, such as:

• Tangential skim
• Exponential skin
• Exponential curve fitting
• Straight line skim
• Front peak skim
• Rear peak skim
• Peak valley ratio
• Valley height ratio
• Valley drop

4. Concluding Comments

The integration of chromatographic peaks is one of the most regularly performed activities in the modern analytical laboratory. In order to obtain consistent results, peak integration must be carried out in the same manner each time the analysis is performed. It is for this reason it is recommended a laboratory establishes appropriate policies and standard operating procedures which provide clear instructions on what constitutes a chromatographic peak and accurate peak integration.

In addition, clear procedures should be established on how the correct chromatographic parameters should be developed and validated during method development and validation. Once a method has been validated the same integration parameters should be used for all subsequent analysis performed using that method. In order to achieve this, it is recommended the approved written analytical test methods contain either a list of method integration parameters, or a list of proven acceptable ranges of method integration parameters.

5. References

1. M. D. Smedley, US FDA Warning Letter No. WL: 320-13-22, US FDA, Silver Spring, MD, Issued 30 Jul 2013, http://www.fda.gov/ICECI/EnforcementActions/WarningLetters/2013/ucm369409.htm
Accessed 3 Jan 2015
2. D. Amador-Toro, US FDA Warning Letter No. 13-NWJ-10, US FDA, Silver Spring, MD, Issued 2 Jul 2013, http://www.fda.gov/iceci/enforcementactions/warningletters/2013/ucm359637.htm
Accessed 3 Jan 2015